Uh … holy shit.

[brianjopek@mastodon.world]

Uh … holy shit.

“What could possibly go wrong?” - The Resistance

[jbaggs@infosec.exchange]

Where is the "follow a bit less" filter?

[felipeb@hachyderm.io]

@BrianJopek
laws are written in blood indeed

[poleguy@mastodon.social]

@BrianJopek The day after the school got hit I called that it was probably an AI behind it. As soon as I saw the satellite image of where the school was it was clear to me.

The ai sees no consequence for getting it wrong.

[gbargoud@masto.nyc]

@BrianJopek

Hate the phrasing "AI was responsible for". If I drew a grid and rolled some dice to pick which part of the grid to bomb, the dice aren't responsible for shit.

[grumble209@kolektiva.social]

@BrianJopek Well, when people die from this horseshit there's a better than zero chance it will be a lobbyist or a congressman or someone who can change the law.

[psh@infosec.exchange]

@BrianJopek before accountability for these systems are even set in place at that...Not to mention the hundreds of overcrowded airports and near misses...I'm sure this will turn out well.

[shadsterling@mastodon.social]

@BrianJopek https://www.cryovex.com/ai-air-traffic-faa-deploys-875-million-tool/

“a collection of machine‑learning models trained on decades of historical flight data”

I dunno, this might not be a misapplication of a general-purpose LLM to something it can’t do, it might be a misapplication of the term “AI” to an ML tool that is actually well-suited to its one and only task

[mimblewimble@mastodon.social]

@BrianJopek if I can get off target responses from the frontier models after a 30 minute discussion I shutter to think what is going to happen when they run for endless hours on a task like air traffic control. I don't know how the Moonshot guys say they can run overnight on an unsupervised task. That's bullshit.

[yakyunightowl@mastodon.world]

@BrianJopek On the bright side, they can't build new things in the city when the old things need emergency repairs from aviation incident debris.

[brianjopek@mastodon.world]

@ShadSterling I guess we’re gonna find out! Hopefully, not the hard way.

[tsvga@ni.hil.ist]

@BrianJopek I see a sequel to the Tenerife airport disaster in the future

[vtdarksim@dmv.community]

@ShadSterling @BrianJopek *potentially* a good application of ML (not LLM), but not under these fucking morons.

[deepmud@mas.to]

@BrianJopek and 911, God help us.

[perfidia@autistics.life]

@BrianJopek Duffy is a lazy good-for-nothing.

[jjdavis@infosec.exchange]

@BrianJopek Hello, Amtrak?

[bcasiello@floss.social]

@BrianJopek
With a high IQ president, that wouldn't have happened.

[ashmire@pagan.plus]

@BrianJopek Apparently they're just determined to find a way to make the claims of AI killing everyone come true one way or another.

[washingtonirving@mastodon.world]

@tsvga @BrianJopek Tenerife was the very example of downsides of human behavior. I am very against AI doing ATC. Just saying it was the worst example

[kaito02@mastodon.social]

@BrianJopek I hope it "misleads" AF1